The problem

We were producing one or two blogs a week, and I knew we were leaving organic growth on the table. The process itself wasn't broken, but it was entirely manual. Every post required someone to research a title, check for content overlap, write the copy, format it for WordPress, and populate the SEO fields before pushing to draft. Each of those steps ate into time the team could have spent elsewhere, and the output simply wasn't enough to compete in search. The other issue I identified was duplicate intent. Without a systematic way of checking what we'd already covered, there was a real risk of writers circling the same topics in different language. Duplicate content dilutes domain authority, and I wanted a way to eliminate that risk entirely rather than manage it manually.

Our approach

I designed and built a fully automated blogging pipeline in N8N, running on a Cron schedule that fires twice a day, Monday to Friday. The system starts itself at 9am, runs again at 4pm, and produces a complete blog post both times without anyone touching it.

Finding a gap in the content

The first thing I wanted to solve was title strategy. Every blog title we'd ever published lives in a Google Sheet, and I set the workflow to call the Sheets API and pull the full archive before passing it to GPT-5 mini. That model carries a 400,000 token context window, which means the entire content library fits in a single API call with room to spare. The model reads the full list, identifies the gaps, and generates a fresh original title we haven't covered yet. No overlap, no duplicates, and a content strategy that gets smarter the more we publish.

Writing the post

That title goes to Claude Sonnet 4.5 or 4.6 or OpenAI GPT5 as a writing brief. Claude handles the full article, producing copy with a coherent argument and natural structure. I chose Claude specifically for this step because the output quality is consistently high enough to publish with minimal editing.

Formatting for WordPress

A dedicated LLM agent takes Claude's output and converts everything into clean HTML, structured correctly for WordPress blocks. The post arrives in the CMS formatted and ready, with no manual reformatting or broken spacing to fix.

SEO metadata

A fourth LLM generates the page title, the URL slug, and the meta description, feeding them directly into Yoast SEO via the WordPress REST API. By the time anyone opens the draft, every SEO field is already populated and scored.

Human review

The workflow creates the post as a draft, Yoast scores the SEO, and a notification goes to the team. A human reviews the draft, checks the score, and approves or edits before anything goes live. The whole review takes about 10 minutes. I kept this step deliberate because quality control matters, and the workflow is designed to support editorial judgement rather than replace it.

The results

Output went from one or two posts a week to five posts a week, every week, with no increase in headcount and no change to what anyone on the team was responsible for.

The compounding effect on organic traffic was exactly what I had anticipated. More indexed pages created more entry points for search, traffic climbed consistently week over week, and that sustained volume of fresh, relevant content led directly to more enquiries and increased sales.

Quality held up throughout. Every post went live with a proper slug, a populated meta description, and a passing Yoast readability score. Volume and quality moved together rather than trading one off against the other.

The efficiency gain was significant too. A post from concept to draft used to require two to three hours of human time. The workflow reduces that to a 10-minute review, returning 10 to 12 hours to the team every week to focus on higher value work.

PS. Don't worry, these case studies are fully human written! :)